Augmented Microscopy for DNA Damage Quantification: A Machine Learning Tool for Environmental, Medical and Health Sciences

Abstract

The Comet Assay is a well-known procedure employed to investigate the DNA damage and can be applied to several research areas such as environmental, medical and health sciences. User dependency and computation time effort represent some of the major drawbacks of the Comet Assay. Starting from this motivation, we applied a Machine Learning (ML) tool for discriminating DNA damage using a standard hand-crafted feature set. The experimental results demonstrate how the ML tool is able to objectively replicate human experts scoring (accuracy detection up to 92%) by solving the related binary task (i.e., controls vs damaged comets).